A fault-tolerant control method based on adaptive FNN for ship control system

This paper presents an adaptive controller based on adaptive fuzzy neural networks (FNN) to implement the fault- tolerant control of the rudder in a ship system, which could avoid the deficiencies of the unmatched model and the imprecise description of the failure pattern in the traditional methods. And an approximate digital differential algorithm with variable learning rate is proposed for the neural network learning. It is particularly suited with undetermined models. In the paper, the principle, the structure and the learning algorithm of the adaptive FNN controller will be discussed. And some simulation results show that this controller has better fault tolerant control ability than PID controller to keep system performance in the partial failure of the actuator.